Suppose we have a small number of very close duplicates on some particular question. Suppose also that the questions are high quality and have high quality answers - at least, as measured by the StackExchange voting system. Should we necessarily close some of the questions as duplicates?
The answer, according to Jeff Atwood's StackOverflow blog post is definitely no. In fact, he argues that some duplication is good, as far as search optimization and expression of multiple viewpoints go.
This question is specifically motivated by the following two questions, the second of which was closed as a duplicate of the first - nearly four years after being asked:
Both questions have a lot of votes and have answers with a lot of votes. The answers have quite a different nature, however. The first one contains a rather encyclopedic list of references on the topic. The second one (that's now been closed) has two answers that play off of one another nicely and is much more self-contained. I think that both questions are quite valuable.
I'm certain that there are other duplicates - here's one, for example, that I think is nowhere near on the same level as the other two. It seems silly that this one is not closed. I would argue that the one that was closed was closed precisely because it is good enough to be visible. That seems odd.
In fairness, I should mention that I authored one of the answers for the closed question. Nonetheless, I'm of the opinion that the practice of closing questions as duplicates is to prevent dozens of duplicates from arising, rather than just a few, and that quality should be accounted for.